Nstreme interfering with my other AP

RouterOS Wireless Networking
1

I’ve got a 5.8GHz 23 mile link (-70s tx and rx) and a 5.3GHz 4 mile link (-50s turned down to -70s)back to back on the same tower. Running 802.11, they both will connect at 54mb with high CCQ, everything is working pretty well.

The 23 mile link will only move around 5mb on a bandwidth test, even when connecting at 54mb. When I enable Nstreme It will push 10mb consistantly. The problem is when I enable Nstreme and stress the system it causes the 5.3GHz 4 mile link to drop to 6 or 9mb with low CCQ.

If I enable Nstreme on the 4 mile shot it will cause another shot coming from that tower to drop entirely. This second shot will be moved to a separate RB shortly.

If I enable Nstreme on both links will it help with my self interference or will it cause problems on both sides?

Any tips or tuning recommendations?

2

I have had many similar interference related problems lately but am solving almost all:

  • Separate working freq’s of close proximite radio’s as much as possible, even if not on the same tower.
  • To get more spectrum space available try to use 10Mhz wherever possible.
  • Physically separate radio in same tower as much as possible.
  • Try not to use 2 or more radio’s in same box. They tend to interfere with eachother.
  • Put routerboards (with only 1 radio) in metal box or shield plastic boxes with alu foil or metal spray paint.
  • Make sure antenna’s are not too close to eachother. If horizontal distance not possible move them vertical.
  • ´Harden´ your link with putting opposite radio in either ´access list´ or ´connect to´ list and mention both SSID and mac address in two separate ´connect to´ rules in station. Make station to scan on AP freq. only.
  • Remove as much variables router can choose on (like preamble, freq’s) as possible. If both sides know exactly what the setting of the other side is less time is waisted in the ´hand shaking´.
  • forget about ´nstreme´. Use NV2 enabled with security. It works now pretty stable in ROS 5.1.
  • Use ROS5.1. It is stable and for normal wireless links I have not seen issues so far. Finally! I am implementing it on all my links and clients now.
  • Make sure that after update of ROS you also update firmware.

Wireless links are the basics you have to work on to start building a network.
It needs carefull planning and testing and fine tunint to get the best.

If you are the only user of the spectrum you have an easy life. But beware if the spectrum becomes filled with many other users! Including your own!

3

Isn’t NV2 going to be less spectrum friendly than Nstreme?

The rest are pretty much already there. This is my third year in the business, got a few customers. Talking to some other guys on here sounds like they’ve had multiple 5GHz shots in the same box with good CCQ. I have multiple shots from separate metal boxes interfering with each other.

Admittedly I don’t have much room on my tower. This was the first tower I put up, funds were tight, so it is just above the tree tops. The top 10 feet is really all I have to play with right now.

4
  • forget about ´nstreme´. Use NV2 enabled with security. It works now pretty stable in ROS 5.1.

I haven’t installed 5.1 yet but don’t see NV2 package unless it’s included in the wireless package.

I have a 22Km link signal varies between 60 to 75db, using 4.17 and NV2 disconnects and reconnects approx every 2 hrs, but using nstreame and 802.11 the link is stable, must try 5.1 but i suspect that some of the other Ap’s and links on this mast could be causing this issue,

I agree with the advice given by WirelessRudy but when you have taken the links+AP’s the max distance apart on a mast or tower, the next step i am considering is to place a “earthed” grid mesh between each of the links, what grid mesh size will attenuate both horiz and vert polarity signals, and hopefully this will reduce the unwanted signals but windloading is another consideration?

5

Use different polarizations for the links. This gives a lot of additional separation.

6

The main feed coming into this tower is horizontal and it doesn’t seem to cause any interference or catch any, so this is definitely a big part of the answer.

If I enable Nstreme on all of my 5GHz stuff, will this likely fix my interference or just completely hammer my entire network down?

7

Maybe yes, I have not so much experience with nstreme. It never worked for me. But I do now have experience with nv2 and yes, it has been showing to have impact on other links. But since the competition is using it (Airmax, just another manufacturers way of using TDMA, which nv2 is) I had to follow to keep my links running.

The rest are pretty much already there. This is my third year in the business, got a few customers. Talking to some other guys on here sounds like they’ve had multiple 5GHz shots in the same box with good CCQ. I have multiple shots from separate metal boxes interfering with each other.

I have had, and still have, several multi ´shots´ in one box. But now I am struggling with several links that are more and more suffering from interferences, whether from my own other links or 3rd party, I decided to fight the whole issue over by starting at the basic. Good radio links which means radio’s also have to be protected as good as possible from other interfering sources.

Admittedly I don’t have much room on my tower. This was the first tower I put up, funds were tight, so it is just above the tree tops. The top 10 feet is really all I have to play with right now.

Don’t we all have that problem? :slight_smile:
I merely try to give you directions in how to setup things. Off course I also can not always create ideal situations. So yes, I do have boxes sharing radios. Yes, I do have masts where antenna’s are almost touching each others. In the ideal world I would be the only provider with nice high masts and the best available equipments for clients that pay me in gold… :smiley:

8

It is.

I have a 22Km link signal varies between 60 to 75db, using 4.17 and NV2 disconnects and reconnects approx every 2 hrs, but using nstreame and 802.11 the link is stable, must try 5.1 but i suspect that some of the other Ap’s and links on this mast could be causing this issue,

5.0/5.1 solved most of my disconnect issues. But, like any other link, fine tuning the link for stability is also under nv2 very important. Play with your TDMA Period Size (not too small!) and the cell radius. Usually nv2 has an influence on ´other´ 802.11/nstreme links, not the other way round. Also check if there are not other sources of interference around. Try to keep the nv2’s link frequency at least 40Mhz away from any other frequency.

[I agree with the advice given by WirelessRudy but when you have taken the links+AP’s the max distance apart on a mast or tower, the next step i am considering is to place a “earthed” grid mesh between each of the links, what grid mesh size will attenuate both horiz and vert polarity signals, and hopefully this will reduce the unwanted signals but windloading is another consideration?

“Shielding” the boxes containing your radio’s should be enough. Preferably use metal boxes. An ´earthed´ grid sounds a bit over the top to me. You also have to consider that the signal the remote stations are sending to your mast (since here is the radio they are ´talking´ to) hits your mast everywhere. Also signals from other band users hit your mast. The only way to do some good shielding against all this is to encapsulate the radio inside a ´faraday’s cage´ so the only signal entering the box is the one brought in by the pigtail connected to your radio.
I would forget about your idea of the big grid mesh. But if you do, be aware that 5Ghz has a wavelength of some 3cm’s. So the raster (or ´holes´) have to be smaller than that.

9

“Shielding” the boxes containing your radio’s should be enough. Preferably use metal boxes. An ´earthed´ grid sounds a bit over the top to me. You also have to consider that the signal the remote stations are sending to your mast (since here is the radio they are ´talking´ to) hits your mast everywhere. Also signals from other band users hit your mast. The only way to do some good shielding against all this is to encapsulate the radio inside a ´faraday’s cage´ so the only signal entering the box is the one brought in by the pigtail connected to your radio.
I would forget about your idea of the big grid mesh. But if you do, be aware that 5Ghz has a wavelength of some 3cm’s. So the raster (or ´holes´) have to be smaller than that.

I agree with you now after swapping out a 24db grid for a 30db grid on a 22km link, using the 24db grid -62 and swing of 16db over 24hrs and hourly nv2 disconnects but was stable on 802.11, now signal is 57db with a swing of just 3db with stable NV2 connection. I wonder if this was due to two factors higher gain = better signal+SNR and narrow TX+RX beamwidth of the 30db grids = reduced area to pickup of noise on the 22Km?

10

I always use ´over the top´ db antennas. Apart for reaching longer distances these antenna’s have the higher gain because they transmit the radio energy in a smaller ´beam´ while when listening they pick up less ´foreign´ signals (=noise) from directions outside the beam. I use them on relative short links. In very short links where my antenna choice might work out to give a too strong signal (-40dBm or even lower) I just lower the radio output. As bonus you save some energy, produce less heat in the antenna casing, create less ´noise´ for other antenna’s/towers in region and your radio will last longer too!

So the extra budget spend on the antenna’s earns itself back in better signals and higher (´lower´!) noise levels.
I also noticed with nv2 that on marginal links where data rates are to be set lower (and nv2 prefers to have only one level set anyway, it better be a stable one) or where signals are in the -70 to -80 range it creates too many disconnects which seems to effect other links in reach!

So yeah, your good choice for the higher gain antenna is working out! :smiley:

11

I always use ´over the top´ db antennas. Apart for reaching longer distances these antenna’s have the higher gain because they transmit the radio energy in a smaller ´beam´ while when listening they pick up less ´foreign´ signals (=noise) from directions outside the beam. I use them on relative short links. In very short links where my antenna choice might work out to give a too strong signal (-40dBm or even lower) I just lower the radio output. As bonus you save some energy, produce less heat in the antenna casing, create less ´noise´ for other antenna’s/towers in region and your radio will last longer too!

So the extra budget spend on the antenna’s earns itself back in better signals and higher (´lower´!) noise levels.
I also noticed with nv2 that on marginal links where data rates are to be set lower (and nv2 prefers to have only one level set anyway, it better be a stable one) or where signals are in the -70 to -80 range it creates too many disconnects which seems to effect other links in reach!

So yeah, your good choice for the higher gain antenna is working out! > :smiley:

What has amazed me is the signal varation from some 16db to just 3db over a 24h cycle,

To date i have only used grids rather than dish’s, windloading and having a very strong tower to support these dish’s is another issue, what is the highest gain available on a grid,

12

FYI …

Card: R52Hn running default power, 10MHz BW channel.

Regards;

13

I’m trying to ´read´ the graph but am left with some questions:
The legenda doesn’t make much sense to me; “min power” (blue) shows power peak (?), Average (Green) is always the same over the whole band exept in nv2.
And what is the yellow “current” line telling me?

The only premature conclusion I think to be seeing is that nv2 station has slightly higher power output while other legacy ´b´ device has less..?

Please explain a bit more.

14

Hi …

Sorry, the pictured was reduced in size to fit the html page with minimum scrolling.

The blue area is not minimum but maximum. And yes, shows power peak. As 802.11-not-N modes are half duplex this filled area tooks some time to build up because the test equipment scans full spectrum ~4 times per second. So the chance to it be “listening” exactly the frequency where 802.11 signal is are small. Several scans are necessary to integrate existing power at certain portion of the analyzed spectrum.

The green line - average - seems like noise floor due to test equipment sampling method. May be NV2 have this “hump” at the - useless for me - average line due to the 2 miliseconds bursts, against 25 microseconds from regular 802.11 half-duplex. So the test equipment have more energy sampled per period of time (see, it’s a Us$ 50 USB device, not a HP or R&S +15k Us$ with the plugin firmware to 802.11 modes lol).

The yellow line is the current sampled signal amplitude. Useless to TDM/TDMA modes but usefull to “watch” some narrow band signal such as cordless phones, steady carriers, local oscilator leaks & etc.

Hmmm … this is an off-air sample, both ~20dBm … different antennas. The correct analisis should be done in a different way, both radios connected to a 2 way splitter/combiner with > 60dB port isolation and the common port connected to the spectrum analyzer. And - again - I guess the peaks on NV2 are introduced by some limitations the test equipment. Unless MT is using 4 of 54 OFDM carriers as pilots or something like that.

The total power is the same … but G modes keep the energy restricted to its 52 (or 54, I don’t remember) carriers. B mode have a single carrier spreaded in the spectrum. Since power is proportional to the filled area, the RMS value match but the peak value will be different (802.11B spreads a bit more).

Anyway, the ony thing I try to show on such pictures is that the “G” modes spectrum use are pretty the same. Received power is ~60dBm while intermod products are below -25dBc (e.g less than -85dBm).

Now, if second card front end does not handle vy well existing power from the first one, all intermod products will raise at this card, due to RX low noise amp / mixer overload. But the irradiated spectrum mask is what the pictures are showing.

Regards;